1:
2:
3: \section 6. Comparison with chiral perturbation theory
4:
5: In two-flavour QCD with unbroken isospin symmetry,
6: the chiral expansion of the pion mass reads [\ref{GasserLeutwyler}]
7: \equation{
8: \quad\mpi^2=M^2
9: +{M^4\over32\pi^2 F^2}\ln(M^2/\Lambda_3^2)+\ldots,
10: \qquad M^2\equiv2Bm,
11: \enum
12: }
13: where $F$, $B$ and $\Lambda_3$ are a priori unknown constants.
14: A phenomenological analysis, taking low-energy experimental
15: data as input, suggests [\ref{GasserLeutwyler},\ref{ColangeloDurr}]
16: \equation{
17: \quad F=86.2\pm0.5\,\MeV,
18: \qquad
19: \lbar_3\equiv
20: \left.\ln(\Lambda_3^2/M^2)\right|_{M=139.6\,\MeV}=2.9\pm2.4,
21: \enum
22: }
23: while $B$ depends on the renormalization scheme for
24: the quark mass and thus cannot be determined from such data alone.
25: The chiral expansion of the pion decay constant
26: has the form [\ref{GasserLeutwyler}]
27: \equation{
28: \quad\Fpi=F-{M^2\over16\pi^2 F}\ln(M^2/\Lambda_4^2)+\ldots\enum
29: }
30: and the phenomenological discussion leads to the estimate
31: [\ref{ColangeloGasserLeutwyler}]
32: \equation{
33: \quad\left.\lbar_4\equiv\ln(\Lambda_4^2/M^2)\right|_{M=139.6\,\MeV}
34: =4.4\pm0.2\enum
35: }
36: for the low-energy constant $\Lambda_4$.
37:
38: In principle the low-energy constants
39: can be determined from lattice data without recourse to
40: phenomenological estimates.
41: However, as will become clear shortly,
42: the available simulation data are
43: insufficient for a solid analysis
44: of this kind. While the situation
45: in the case of the pion mass is somewhat more favourable,
46: our principal goal in the following will be to find out
47: whether the data are compatible with the expansions (6.1) and (6.3)
48: for a reasonable choice of the parameters.
49:
50: We first need to rewrite the equations in a form where all dimensioned
51: quantities are expressed in units of the scales at the reference point.
52: To this end, it is helpful to introduce the abbreviations
53: \equation{
54: \quad
55: x={2m\over\mref+\msref},
56: \qquad
57: C={\MKref^2\over32\pi^2\FKref^2},
58: \enum
59: }
60: and to define the scaled parameters
61: \equation{
62: \quad\Fhat={F\over\FKref},
63: \qquad
64: \Bhat={\mref+\msref\over \MKref^2}\,B,
65: \qquad
66: \lhat_n=\ln(\Lambda_n^2/\MKref^2).
67: \enum
68: }
69: The chiral expansions
70: \equation{
71: \quad{\mpi^2\over\MKref^2}=\Bhat x
72: +C{\Bhat^2 x^2\over \Fhat^2}\bigl\{\ln(\Bhat x)-\lhat_3\bigr\}
73: +\ldots,
74: \enum
75: \next{2ex}
76: \quad{\Fpi\over\FKref}=\Fhat
77: -2C{\Bhat x\over \Fhat}\bigl\{\ln(\Bhat x)-\lhat_4\bigr\}
78: +\ldots,
79: \enum
80: }
81: may now be directly compared with the simulation data
82: (note that $\lhat_n=\lbar_n-2.53$ if $\MKref=495$ MeV is assumed).
83:
84: \input figure6
85:
86: The computation of the decay constant $\FKref$,
87: and thus of the constant $C$, involves
88: the renormalization constant $\ZA$ of the axial current.
89: Recent estimates of the latter in the two-flavour Wilson theory
90: at the couplings of the $A$ and $B$ lattices are
91: $0.77(2)$ and $0.78(2)$ [\ref{BecirevicEtAl}], while in
92: the case of the $D$ series of lattices we may use the value
93: $0.75(1)$ determined by the ALPHA collaboration
94: [\ref{DellaMorteEtAl}].
95: For the constant $C$ we then find
96: $0.068(4)$, $0.071(4)$ and $0.076(3)$
97: respectively.
98: These figures are barely consistent with one another,
99: suggesting the presence of lattice or finite-volume
100: effects, but it should also be noted
101: that the determination of $\ZA$ is not free of systematic ambiguities
102: [\ref{BecirevicEtAl},\ref{DellaMorteEtAl}]%
103: \kern2pt\footnote{$\dag$}{\footnotefont%
104: The same comments apply in the case of $\FKref$
105: where we obtain $107(3)$, $105(3)$ and $101(2)$ MeV
106: on the $A$, $B$ and $D$ lattices
107: (assuming $\MKref=495$ MeV as before).
108: The fact that these results are
109: all lower than the decay constant $\FK=113(1)$ MeV
110: of the physical kaon may not be significant,
111: because the two-flavour theory neglects the
112: effects of the strange sea quark.}.
113:
114: \input figure7
115:
116: A very accurate determination of $C$ is fortunately
117: not needed for the chiral fits,
118: because $C$ only appears at next-to-leading order
119: in the chiral expansions.
120: We thus set $C=0.072$ and simplify the fit procedure
121: by substituting $\Fhat=0.70$ in eq.~(6.7),
122: which will turn out to be an approximately correct value.
123: In the range $\Mpi/\MKref\leq1.1$,
124: the one-loop formula (6.7) then fits the data for the pion mass very well,
125: the fit parameters being
126: $\Bhat=1.11(6)(3)$ and $\lhat_3=0.5(5)(1)$ (see fig.~6;
127: the second errors are
128: estimates of the systematic uncertainty arising from the inaccurately
129: known values of $C$ and $\Fhat$). We did not attempt to estimate
130: the effects of any higher-order terms in eq.~(6.7) so that the
131: quoted values of the fit parameters
132: should be taken as effective values, describing the data in
133: the specified range of pion masses.
134:
135: In the case of the pion decay constant,
136: the comparison of the simulation data
137: with the chiral formula (6.8) is complicated by the
138: scattering of the data points in fig.~5, which may partly
139: be the result of systematic effects.
140: However, since the points line up at the smaller
141: quark masses, we may attempt to fit these,
142: setting $C$ and $\Bhat$ to the previously determined
143: values and adjusting $\Fhat$ and $\lhat_4$
144: (see fig.~7).
145: The statistical quality of this fit (solid line) turns out to be quite good,
146: but the curvature of the fit function is not seen in the data and
147: the fit therefore appears to be somewhat artificial.
148:
149: A more plausible fit (dashed line)
150: can be obtained
151: by including a hypothetical two-loop term proportional to
152: $\Bhat^2x^2/\Fhat^3$ in the chiral expansion (6.8), with a
153: coefficient $C'=0.046$ that is not unreasonably large.
154: The fit parameters $\Fhat$ and $\lhat_4$ change
155: from $0.60(4)$ and $1.6(1)$ to $0.73(3)$ and $0.73(8)$, respectively,
156: when the two-loop term is added.
157:
158: The discussion in this section shows that
159: simulation data at significantly
160: smaller quark masses, with small systematic and statistical errors,
161: will be required for a reliable determination
162: of the parameters in the chiral lagrangian.
163: It seems safe to conclude, however,
164: that our results in the range $\Mpi/\MKref\leq1.1$ are not
165: incompatible with chiral perturbation theory.
166: In particular, the fact that $\Mpi^2$ is a nearly linear
167: function of the quark mass $m$ is not in conflict with the
168: presence of the chiral logarithm in eq.~(6.7).
169: